 All right. So I just have five slides, so it should only be maybe 45-50 minutes. So it's a little bit about the breakout session yesterday in pharmacogenomics, and Alan Shuldiner wasn't here, but this is on behalf of the group that met, including the three that have some responsibility. This was our charge from last December, identifying one or more collaborative demonstration projects that would advance the implementation of pharmacogenomics into clinical practice. And I'll mention briefly one of them in the next slide, but very little time was spent on that, mainly because the RFA is out, people are already thinking about those actual projects from the past discussion, and if you haven't started thinking about it now, it's kind of too late, so there wasn't a lot of time spent on this part of the charge. Now this is a slide from Alan. I'm not sure exactly what it says, so I'll just talk and not even look at it. But basically one of the things that is coming out of the last efforts is this collaboration that has been heavily driven by the Pharmacogenetic Research Network, where many of the PGRN investigators, including Julie, Dan, others who are probably here that I'm not seeing at the moment, have been involved in getting their version of the must-have genetic variants. Debbie Nickerson, who was here yesterday, is involved in putting together that platform. And there's a custom drive panel, the VIPGX or PGRN-Seq or whatever it ends up being called, will be applied probably across some of the eMERGE network sites, which also is heavily represented here in the room. And so the idea that a more focused platform for looking at pharmacogenetic variants can be tested in a broad population, looking at what is the incidence of these variants, looking at some of the initial association with outcome, is part of the idea that's coming forward. And this is mainly driven because many of the off-the-shelf platforms do not do a great job. Many of the off-the-shelf whole exome, next-gen, sequencing type platforms do not do a good job when it comes to pharmacogenomics. And I think many of us have had experience using the several of the different next-gen platforms, fantastic coverage, except for SIP2D6 and the 3A family, and some of these other regions where there's a lot of pseudogenes, high homology, and those kind of issues. And so one part of that is really trying to tackle this. Now, I think that some of the off-the-shelf, if you want to use that term, tools that are coming out now are starting to overcome this issue. There's also other labs that have their own custom version of this that seem to give very good coverage. So I think this is a problem that technically is being solved. But where this comes in is really jumping the hurdle towards doing a broader application, taking these platforms and trying to think about what is the instance. You know, when I talked to, I love that Pearl had such a primary care focus to a lot of what she was just talking about, because when I talked to family members who are in the primary care area or collaborators for that matter, first of all, they want to know just in time. But they also want to know what are the areas where they should care. It's really kind of what is the minimum amount of information that can be delivered as late as possible. And I think some of the work that's happening here will help identify things like how many patients actually have these variants. You know, is this a one-in-a-lifetime experience or every other patient? I don't know if Josh Jean is still here, but yeah, right there. There he is. Yeah, you're a pair. Congratulations. So I know you got a paper that will be coming out sometime soon. The editor of the journal is back there, so maybe I shouldn't say anything. It's probably an embargoed. But just looking at, well, what are the frequency of some of these variants? What are the frequency of some of these variants in a Vanderbilt medical home? And it actually occurs in a lot of people, and it's not this rare event. And I think some of that, some of what's coming out of this pharmacogenomics working group, are getting some of the simple descriptive proof of principle that would cause a generalist, whether they're an internist, family medicine, pediatrics, to realize that they might actually care someday. And they can wait until they have to care, but they at least might care someday. And so I'll stop rambling about that. The other thing is we spent a lot of time on non-RFA issues because the reasons I already said. And one of them was around the areas of publication. And so one of the things that really came out somewhat from Ned's talk yesterday and from also general discussion is this idea that in pharmacogenomics there are elements that are not usually included in genetic assessment analysis. And we talked some about that. I mean, it's not just EGAP. It's really across the board with, I guess, one of the exceptions being Mary's CPIC program. But the idea of going and looking at across these, doing kind of a, Mark called it a meta-meta-analysis, a meta-analysis of the meta-analysis, is not really that. What it is, is basically assessing the different assessments and saying, well, what are some of the categories of information that are consistently missing and try to inform the assessment community about those? Because there are just a small number of categories, things like dose or concomitant medication, that if the EGAPs and others knew about them, they would become part of the normal process and would allow the interpretation of the pharmacogenetic component to be a bit clear. The answer still may be that they're completely useless. The test, not the assessors. But the idea that we can do this and have an outcome that would cause people to maybe be disgruntled but not have a good reason would be ideal. Because right now, you get a lot of feedback about disgruntled and they use this as an excuse. It would be nice to go through that. The other, and really so we can just define some of the rules to aid these groups. The second is looking at some of the advantage and disadvantage of the current platforms for pharmacogenetics. And there are a number of platforms out there. This is something maybe done best with the sequencing group. There are a number of platforms out there and we're starting to learn a lot about the nuances of where they're useful and where they're not. And there are a lot of general laboratorians who maybe do molecular diagnostics but also do a bunch of other stuff at the mid to small size centers that are starting to get into the space. They're taking off the shelf products and hoping doing the best they can. Now, maybe the point should be they shouldn't do that. But they're doing it anyway. And so we could come up with some output that would help put them into context. Because there are some major differences across these platforms when you get to the practical side of that. And we talked a lot. How is it was was was was kind of to join our group and get all the glorious benefits of being in our group last night. We had the free car and all that all the booze. And really was highlighting some of what's going on at LabCorp where they have kind of every platform possible and have to figure out which one to use in each case. And many many times that is the situation. The last thing is that amongst amongst the different genetic areas there there often is a a single or a small number of sugar daddies that are pushing a particular area might be inherited metabolic disease might be in cardiology etc. But in pharmacogenomics there really isn't a body that way. And if you expand it further and look at imaging, imaging happened rapidly because there were two bodies that were involved. One was the manufacturers making sure that everyone was trained up on how to use the latest scanner had access to it had experience to it. And the second was the the radiology is a discipline that exists and say depended. It's a discipline to depended on to translate complexity into simplicity. And genetics does not have that that in most centers. And so this this idea of trying to really look through who should be pushing this, including trying to engage the payers a little bit more. Because when it comes down to it, the payers are the ones who are likely to benefit first, even before the patient in in many of the things we're talking about. And then lastly, we really highlighted some of the endpoints that we were interested in academically. And they often don't match what insurance companies care about, and certainly don't match what even our own clinical leadership might care about within our institution. You know, it was wasn't until meeting with our CEO that I realized how important bounce back in the first 30 days was. That was not an endpoint on my radar. It is now. The you know, half a day of delayed discharge for because adequate pain control huge deal. And also, I think it was visiting some of the folks at Cincinnati, one of their colleagues is anesthesiologist, and he mentioned that 22 minutes in the recovery room was equal to $1,500. And you know, minutes matter in that context. So we had talked about going amongst the institutions involved at this, either at this meeting or just with that committee, and starting to define those endpoints better and getting them out in the public domain. Because certainly most pharmacogeneticists are looking at academic endpoints and because that's what study sections recognize. But often they're endpoints that are really not very useful to the people downstream who need to be using it. And then I'm going to end this slide of Rex's barriers for genomic medicine. These are basically copied down from your slide yesterday Rex. And just highlighting this point that many of these things on here are not primary endpoints to our academic studies. And part of that is a education of study sections and selection of study section members. But you know, if you really right now wanted to go for funding around areas that really matter, it's very hard to get this funding in any of the study sections that I've been on. And those of us that are on council, when we think about the fights that would happen at council, with some of the people who aren't in this room right now, because they don't care about things this far downstream, they would be immense. And so there is a challenge both in terms of academics and the process at NIH in terms of trying to really tackle this. And then I'll stop and take questions. Okay, Rex, you first. You're going to argue with the last slide? No. Okay. It was literally his slide. I mean, I just marked it down. So Mark is going to argue with the last slide. No, actually, I'm going to say this is something that I think we'll be coming back towards the end of the meeting in terms of this whole issue of what are the outcomes that we need to be looking at. And I think that's a really important point. And I'm glad that it arose in another setting then in the planning committee setting. I think the other thing, you know, we've been focused a lot on payers, but you know, the point that you made about going to your institutional leadership, it's important not only from the perspective of them as your institution, but also remember the fact that the people that really will benefit from making breakthroughs here are the employers, because they will benefit not only from reduced healthcare insurance premiums, but they also have to deal with absenteeism, presenteeism, all these sorts of issues that affect worker productivity. And then the studies that have been done looking at self-insured plans by ComEd here in Chicago in Illinois, their return on investment for things that work is about five to six dollars for every dollar invested. And so I think that's another community that we should think about when we think about payers is because almost every large company is self-insured. So they are in fact their own insurance company and they stand to benefit and some of them are very interested in exploring new ways to do care. The MedCo studies that we've seen so far have all been paid for by employers, not by MedCo. They get all the credit, but they get the employers to pay for it and because they believe there's something there. Mark. I don't know. I want to comment on Rex's slide. Okay. Okay, thank you. And that is that given the definition of genomic medicine that Terry presented yesterday, in other words, using that information, genomic information in direct patient care, those barriers all make sense. I think other people have a broader definition of genomic medicine and if in that case one of the things that's missing would go right up at the top and that is lack of information about the underlying biology of disease and the ability to use that information to design better therapeutics, better strategies. So I think from at least my point of view that when we talk about implementing genomic medicine, I think we really need to be very clear that we're all agreeing on the same definition because I think a lot of the issues that the disagreements that come about are probably this definitional basis. No, I think that's a good point, Mark. And I think certainly I wasn't trying to, my apologies, Rex, I wasn't trying to get in trouble. I use this as just kind of, I was just trying to anchor with the kickoff of the conference, but the way the slide from you and Eric's paper is it's those five boxes, the graphic artist tried to make them a continuum. And if we end up with people working in each of those boxes, we really haven't gone forward. We need to be people working across those boxes. And that's the challenge is that as humans we'd like to be in a box for some reason and we need to be looking across there. And so if we don't have good discovery, if we don't have good validation, including in some of these clinical trial samples, if we don't have biological plausibility at the least, if not actual mechanism, then you're right. We can't drive this forward. And we can't get any extra added value from it. I just want to add to that. I mean, Mark and I have talked, and I'm used to being in trouble, so no problem there. But one of the things that strikes me that we don't emphasize enough to build on Mark's point is all of these variants of unknown significance that we see in these sequencing projects are likely to be highly informative in terms of the biology at some point. And so we really need to be making sure we're not just moving left to right in that diagram, we need to be moving right to left in that diagram as well. So I think there's a really important opportunity for us to figure out how, and we've emphasized at this meeting a lot, the need to capture these variants of unknown significance and then feed them back into the biology. So I think that's a really important point. Julie, and then... So Howard, the one thing that you didn't raise that I think we talked about and that Mary tried to bring up yesterday is that the idea that pharmacogenetics is a bit different than the disease genetic stuff and that, for example, putting a group of genetic data into a medical record as it relates to pharmacogenetics, none of the actionable variants for which really are associated with disease is probably not gonna have the risk of leading to additional diagnostic tests and workup. And so it really does seem like a better starting place to begin to get things into the record in a preemptive way because it's very unlikely to kind of have that extra workup sort of downside potential negatives because they really are specific to drug response. Well, thank you very much for that. I put that X down at this bullet here because I knew I forgot some stuff, and I was hoping that people would fill in the X. So thank you very much, yes, that we just have a discussion. Scott, and then John. So I just wanted to briefly explore your concept that some of your favorite sites in the genome are not accessed well with, say, exome, but that they will be accessed well with targeted genotyping platforms. So I mean, if you think about it, you mentioned processed pseudogenes and maybe repetitive, difficult areas. I mean, some of the reasons why mapping might fail in an exome study or why capture might fail, I would think would give you trouble in some of these targeted genotyping platforms as well because they involve oligonucleotides. Some of the cheapest platforms are done on whole genomic DNA. Yeah. Well, I think that, so I would look to someone else to give the details about the specific output that Debbie and others have generated, but I think the difference has been the use of a general algorithm across the entire genome versus almost hand-crafting the probes used for the selection tool for these. So one can get around these pseudogenes if you design the probes appropriately, but if you use just a general genome-wide analysis or probe generation strategy, you end up falling into trouble. And so they're all solvable areas, but not if you don't take into account the pseudogenes and such. So the- But presumably you could give that feedback to the manufacturers of exomes in here. I don't know what platform, I assume it's an aluminum platform that Debbie's using because she tends to go that way, but it's not like some magical new method is being invented. It's just the selection tools are more carefully crafted for this specific indication. So I didn't mean to overstate it as some amazing new breakthrough that we're all gonna wanna run out and buy stock in. Or sell stock in. John? Is it still the orientation of the pharmacogenomics group that you have the opportunity for large effects, high odds ratios instead of the situation in common disease where you have small effects or rare, small effects for common variants or large effects for rare variants. And in the pharmacogenomics space, you have ample opportunity for large effects for common variants because there's no previous evolutionary selection that's ever occurred for these environmental compounds that we've never seen before as a species. So I think that that has been the wish for pharmacogenomics. The hope is that it would be easier than the disease genetics, for example, because of that. And there are some examples. The most dramatic are some of the HLA's with these odds ratios of 2,500 and such. You don't even need a statistician or genotyping. You just throw it on the wall and there it is. But the reality is that there are many examples coming out now which have an odds ratio of two. And maybe when you add them up and put in some clinical factors, suddenly you get some prediction. So I don't think it's going to be a case where pharmacogenetics will have odds ratios of 10 and disease genetics will have odds ratios of 1.1. But the other thing is that certainly Mary's published recently on this as of others that there are rare variants that matter. And at least the most recent paper that I know from Mary was with one of the SLC transporters saying that there were rare variants that did contribute to the effect of methotrexate. Now there was an effect of some common variants too. So it wasn't an all or none. But I think that most of the lessons that we're learning from the disease genetics people are being found in pharmacogenetics, even though we wished it was going to be simpler. Does someone else from the committee? I'll just say that for many of the drug response phenotypes we just don't have the data yet. Well, you know, that's a, you know, I presented at the NIH in their, I can't remember what your, the genetics course was, but what it's called last year. And before that did a little analysis of the NHGRI GWAS catalog. And it was 4% of the GWASs that had been done at that point had any drug endpoint at all. And it was only a minor percentage of those that it had more than 500 patients in the whole thing. And so if you look at the criteria that we would recognize as being useful for GWAS or other discovery approaches, and look at what's in the catalog for pharmacogenetics, there's very little. Now, of the ones that are there, they've had dramatic impact. I mean, you know, the IL-28B being one of them. But it's almost like pharmacogenetics hasn't really started in some ways. Of the top 200 drugs, it's only about, I think it's about 15 of them that it had any pharmacogenetic analysis at all. Much less negative or positive. So it's a long way to go. Howard, I'm sure you're working to fix that. No, it'll be available to fix it first. It's probably instructive. So I think as many people in the room know, the eMERGE sites are in the midst of trying to think about genetic variants that they would actually measure in an experimental way, put back in electronic health record and test what the implications are for outcomes or for at least process in terms of that. So at least at the Northwestern site, we've been very actively engaged with our physicians group that's gonna be implementing our genetic variants. It's a general internal medicine crowd that's very engaged and interested in quality improvement. And it's really interesting to me that overwhelmingly the things that they are excited about doing are pharmacogenomic variants. And the things that they're less interested in doing are disease risk scores, common disease risks with relative odds ratios of 1.05. But the pharmacogenomic variants, they're quite enthusiastic about. And Dan will appreciate this. We even reached out to some of our interventional cardiologists and they were very excited about actually doing this. That's because they've heard from their friends at Vanderbilt that it's not gonna kill them to do it. But the other piece of the implementation stuff that we've been thinking about in eMERGE is not disease prediction. So nobody, I mean, we've heard that this morning already, but if you have disease X, what's your outcome likely to be? So if you have hypertension, are there predictors of increased risk for renal failure? If you have diabetes, are there increased prediction for ocular complications? And we think that that might have, number one, increased odds ratios, and number two, might catch the attention of the provider and the patient more than telling some 20-year-old that you're at increased risk based on some genetic marker of type two diabetes when they're sort of sitting in your office eating French fries or something. I guess I give myself the last word. Just one, if I could make a comment to your last one. Although the increased risk is certainly important, the payers will be looking for the next step, which is does the intervention and paying for extra drug, et cetera lead to an improved outcome before they'll want to pay for that. Okay, that's a really important comment. We'll stop there. So we're not really on time, but we're sort of on time. So let's have a 20-minute break. We'll come back at 10.50 by the clock on my computer and then we'll have the short updates from Howard Murray and who's the third person? Jeff. And then we'll get back on schedule sort of.